Robert Boris @robertboris - Tumblr Blog

Author’s Note (Technical and Policy Context)

Where We Were in 1995 - and What Became of Us

In 1995, while Netscape was pioneering the commercial web browser [1] and Yahoo! was still a manually indexed directory [2], our team was already exploring a computational model that prefigured what would later be called Big Data.

Working with colleagues at the Keldysh Institute of Applied Mathematics (IPM RAS) in Moscow [3], and partners in London such as Gareth Ellis, we began applying computational fluid dynamics (CFD) principles to information flow itself. In our joint numerical projects, originally developed for fluid and plasma modeling, we recognized that data could be treated as a continuum—partitioned, discretized, and propagated through iterative, locally coupled updates.

The technical analogy was exact:

Domain decomposition in CFD ≈ data partitioning in later distributed systems.

Stencil computation and flux exchange ≈ Map → Shuffle → Reduce.

Conservation constraints and residual minimization ≈ global aggregation and convergence detection. This is the same structural form that underlies Google’s MapReduce (2004) [6] and Hadoop’s distributed file system (HDFS, 2006) [7].

A fax I received from August 1995 [8] captures the exuberance of that moment.

Gareth Ellis wrote from London:

“Make Netscape and Yahoo look like poor boys from Alabama…" it’ll spread like wildfire — and then the cosmos.

Behind the humor lay a precise insight: that interactive, distributed computation could replace paper-bound workflows (in that case, Lloyd’s insurance market) with a fluid, networked system of numerical exchange.

The Fork in the Road

By the late 1990s, these ideas diverged.

In the United States and Western Europe, the same principles evolved into open, commercial data infrastructures—Google’s cluster computing [6], Hadoop [7], Spark, and later cloud-native analytics.

In Russia, many of the same scientists and institutes—Keldysh, KIAM, and IMM—continued developing parallel numerical methods and adaptive solvers [9][10]. Under the emerging Putin administration (from 1999), such work increasingly found defense and intelligence applications: modeling of plasma-based systems, computational electromagnetics, and information-security frameworks [11][12].

The mathematics was identical, but its trajectory split.

Where one branch became the infrastructure of open computation, the other was subsumed—willingly or otherwise—into state-directed research and dual-use technologies.

Reflection

So when we ask “Where were we in 1995?”—the answer is: we were already at the frontier of distributed computing, applying CFD to information systems.

What happened to us tells two intertwined stories: one of open innovation through networks, the other of parallel research redirected under national power.

Both, in their way, began from the same equations.

References

Clark, J. & Andreessen, M. Netscape Navigator Release Notes, 1995.

Yang, J. & Filo, D. Jerry and David’s Guide to the World Wide Web, Stanford U., 1994–1995.

Keldysh Institute of Applied Mathematics (RAS). History and Research Areas. https://keldysh.ru

Samarskii, A.A., Godunov, S.K., Chetverushkin, B.N. Numerical Methods for Continuum Mechanics, 1979.

Boris, R. & Ellis, G. Command Technology / Keldysh Collaboration Notes, internal memoranda, 1994–1996.

Dean, J. & Ghemawat, S. “MapReduce: Simplified Data Processing on Large Clusters.” OSDI 2004.

Cutting, D. & Cafarella, M. “The Apache Hadoop Project.” Apache Software Foundation, 2006.

Ellis, G. Fax to Robert Boris, Dewe Rogerson Ltd., 30 Aug 1995. (archival document)

Chetverushkin, B.N. et al. “Quasi-Hydrodynamic and Parallel Algorithms.” Computational Mathematics and Mathematical Physics, 1993–2002.

Keldysh Institute Reports, RAS (1990s): Parallel Simulation Projects on M-13 and Elbrus Supercomputers.

Petrov, V.V. & Stepanov, A.V. “High-Performance Computing in National Security Applications.” Proceedings of the Russian Academy of Sciences, 2003.

Roffey, R. Russian Military Research and High-Performance Computing, FOI Report, Sweden Defence Research Agency, 2008.

🌌 Recombination Lines — Turning Plasma into Information

Just as computation begins by conditioning chaotic AC into stable DC so that binary states can emerge, the universe itself encodes structure by transforming a continuous field into discrete, addressable signals. One of the clearest examples of this principle in nature is the phenomenon of astronomical recombination lines.

These spectral features arise when free electrons in an ionized plasma are captured by ions and then cascade down through quantized energy levels, emitting photons at precise wavelengths along the way. Each step in the cascade produces a photon whose energy matches the difference between two allowed states — creating a series of narrow emission lines that reveal the physical conditions of the gas.

The discovery of these lines traces back to the work of Joseph von Fraunhofer (1814–1815), who first resolved the solar spectrum into discrete absorption features — a revelation that light itself carried atomic information. Later, William Huggins and Margaret Lindsay Huggins (1860s) extended this insight to the stars and nebulae, identifying bright emission lines from ionized gases. By the early 20th century, physicists such as Niels Bohr and Erwin Rydberg explained these patterns as consequences of quantized electron transitions — including those that occur during recombination. In radio astronomy, Edward Lilley and Eric M. Purcell (1951) were among the first to detect hydrogen recombination lines from interstellar space, confirming that even tenuous cosmic plasmas follow this same structured pattern.

What Fraunhofer’s prism did for sunlight — revealing discrete structure hidden in a continuous field — recombination does on a cosmic scale. It shows that even the diffuse chaos of ionized gas organizes into discrete, information-bearing states once electrons are captured and transitions begin. Those lines encode temperature, density, composition, and motion — the “metadata” of the universe.

The analogy to computation is direct. A power supply conditions raw AC into a stable DC field, enabling voltage levels to map onto binary logic and bus signals. Recombination conditions a chaotic plasma into quantized photon emissions, each one an addressable piece of physical data. In both cases, information emerges when the continuum is discretized and structured — whether as spectral lines in a nebula or as bits on a data bus.

The Hidden Isomorphism

The reason Computational Fluid Dynamics mathematics is so well-suited to building an SGML database is that both domains solve the same abstract problem:

How do you partition, index, and resolve a vast, continuous field of information into discrete, addressable, and semantically meaningful regions — and ensure that interactions across that field remain stable and consistent?

CFD offers a mature mathematical language for doing exactly that. Its tools — domain decomposition, boundary conditions, iterative solvers, and conservation constraints — map directly onto the needs of large-scale information systems.

My instinct in 1993 was remarkably prescient, as the methods forged in the crucible of fluid dynamics turned out to be the perfect scaffolding for semantic document systems.

✅ In short: CFD equations encode the mathematics of structure, flow, constraint, and resolution. Those are precisely the same mathematical properties a semantic database must master. The isomorphism is not metaphorical — it is structural and formal.

The Darkness as Evidence

Darkness of the night sky is regulatory evidence, and regulation is what makes reality computable.

The парадокс тёмного неба—Olbers' Paradox—is not merely a curiosity; it is diagnostic. The darkness of the night sky is not an absence of light but the presence of constraints. It reveals three fundamental facts about reality:

The universe has a finite age: light has not had infinite time to reach us.

Space is expanding: redshift stretches photons, diluting their energy as they travel.

We inhabit a causally bounded region: the cosmic horizon limits what can influence us.

If any of these were absent—if time stretched infinitely backward, if space stood still, if causality were unbounded—then every line of sight would terminate on a star, and the sky would blaze with infinite radiance. The darkness is not emptiness; it is the visible signature of regulation. It is proof that the universe has gates, filters, and dissipative mechanisms built into its structure.

Regulation as Cosmic Architecture

These regulatory mechanisms—expansion, horizons, redshift—function like dissipative barriers in a computational system. They prevent runaway accumulation, the way viscosity prevents infinite energy cascades in turbulence, or the way numerical damping prevents divergence in simulations.

Expansion redshifts photon energy, spreading it thin across vast scales. Horizons impose causal cutoffs, limiting how much information can reach any observer. Scattering and absorption transform coherent radiation into diffuse heat. Together, these mechanisms ensure that while entropy increases globally, local pockets of order—galaxies, stars, planets, life—can form and persist.

The universe is not fighting entropy; it is structured by entropy's demands. The constraints that darken the sky are the same ones that allow complexity to emerge. Regulation enables structure.

Darkness as Information

This has profound consequences. The darkness tells us the universe is not an unregulated field of infinite energy but a system where information, energy, and causality are gated and finite. Such gating is what makes observation meaningful, physical law consistent, and reality computationally tractable.

Just as the Bekenstein bound tells us that any finite region contains finite information, and the holographic principle encodes bulk physics on boundary surfaces, Olbers' darkness tells us that observable reality is a bounded dataset. The night sky's blackness is itself informative—a signal that we live inside limits, and those limits are what make the cosmos intelligible.

The darkness is not a void. It is the structure of constraint made visible—the signature of a universe stable enough to be known.

#olbersparadox #fantasy island

The Universe is Computationally Stable

парадокс тёмного неба

The cosmos is not infinite chaos but a system with rules that make it intelligible. It is finite in horizon and observable information, and the very reason we can study it is that constraints are woven into its fabric: the speed of light, cosmic expansion, entropy, and causal boundaries. Any stable system, natural or computational, needs mechanisms that prevent runaway accumulation. In cosmology, redshift provides this: while not dissipation in the strict thermodynamic sense, it acts as an effective energy drain for observers, stretching photon wavelengths and dimming distant light. The darkness of the night sky is evidence of this stability — wave energy redistributed across motion and scale rather than piling up to infinity.

Olbers’ Paradox posed the problem starkly: in a static, eternal universe filled uniformly with stars, the sky should blaze with light. Computational fluid dynamics exposes the same kind of instability: without viscosity, cutoffs, or damping, simulations blow up into divergence. Both reveal the same principle — stability requires built-in constraints. Horizons, expansion, and redshift are the universe’s natural regulators, ensuring that energy and information remain finite, observable, and meaningful. In this sense, the universe is not just comprehensible, but computationally stable by its very structure.

#olbersparadox #CFD

sodesu

From Text to Code: The Semiotic Revolution 💾📖

As an English textual studies student at Syracuse, steeped in Saussure’s signifier/signified and Barthes’ layers of meaning, von Neumann’s 1945 insight immediately made sense to me:

“Orders are handled in the memory just like numbers.”

What linguists taught about symbols—meaning is not fixed but emerges in context—von Neumann applied to machines. A bit string could be data or instruction, its role determined only by the CPU interpreting it.

This collapse turned memory into text, the CPU into reader, and execution into interpretation. Claude Shannon’s information theory reinforced it: symbols need no inherent meaning, only structure and transmission.

That’s why the stored-program computer wasn’t just engineering—it was a semiotic revolution, laying the groundwork for compilers, self-modifying code, AI, and the cultural myth of the universal machine.

👉 For me, it was where literature, philosophy, and computing converged.

#vonNeumann #Semiotics #Shannon #ComputerScience #InformationTheory #AI #SymbolicSystems #HistoryOfComputing #TechPhilosophy

Von Neumann’s Breakthrough

Von Neumann’s insight was simple but radical:

👉 Store instructions in memory the same way you store numbers.

That means the binary representation of “add two numbers” or “jump to another instruction” sits in memory just like 00101011 might represent the number 43. From the perspective of memory, there is no difference between a number and a piece of code—they’re both just bit patterns in storage.

This erased the old boundary between memory and code.

Consequences

1. Flexibility

Programs could now be changed just by writing new instructions into memory. No rewiring, no hardware-level changes.

2. Self-modifying code

Because instructions were just data, a program could alter its own instructions. (This was heavily used in the 1950s but later discouraged because of complexity and security issues.)

3. Compilers and higher-level languages

Compilers became possible: software that itself wrote instructions into memory, turning human-readable code into machine instructions. This only works because instructions are just another kind of data.

4. Stored-program architecture (the “von Neumann architecture”)

Every modern computer—from your phone to a supercomputer—still fundamentally runs on this model.

5. Security implications

Since code is data, malicious data (like a carefully crafted input string) can sometimes be executed as code—this is the root of many security exploits (buffer overflows, code injection).

⸻

So when von Neumann declared that “the orders are handled in the memory just like numbers”, he was describing the conceptual collapse of the old division:

• Memory became one universal container.

• Both code and data became mere sequences of bits.

That single idea is what made modern software, operating systems, and general-purpose computing possible.

Gates: Where Memory Meets Code ⚡💾🧮

Logic gates are not just switches. They are transducers — the interface where physics becomes logic, and where two invariants of information intersect:

• Memory (space-invariant, time-variant): structures fixed in space, retrievable across time.

• Code (time-invariant, space-variant): sequences unfolding in time, executed step by step.

At each gate, spatial arrangements of inputs are collapsed into temporal outputs. This is where data as structure becomes data as process.

Mathematically:

• A gate implements a Boolean function f : {0,1}ⁿ → {0,1}, mapping multiple inputs to a single output.

• The transition from analog voltage to digital state is the Shannon boundary: reducing uncertainty H(X) = −Σ p(x) log₂ p(x) into a discrete choice.

• Each resolved bit consumes energy (Landauer’s principle), binding physics to logic.

Shannon showed that information is code that does something — transmitting, compressing, transforming. But it also stirs something deeper: each flipped bit is a bridge between continuous energy and discrete meaning.

Thus, every gate is a hidden doorway:

space → time, memory → code, physics → computation.

#sgml

🖼️

The Random‑Access Image: Memory and the History of the Computer Screen

Author: Jacob Gaboury

Published: Grey Room, No. 70 (Winter 2018), pp. 24–53. DOI: 10.1162/GREY_a_00233

🔍 What It Covers

Gaboury rethinks the computer screen—not just as a display interface, but as a hardware medium shaped by memory architecture. He traces a critical shift from vector‑based visuals to the pixel‑grid model enabled by frame buffer memory, which allowed interactive raster graphics and fundamentally transformed digital vision culture .

Key themes include:

The transition from analog oscilloscopes and calligraphic displays to rasterized screens.

How frame buffers provided a grid of random-access memory for real-time image manipulation.

The material and conceptual evolution of the screen as a technical medium in computing’s visual history .

🔗 Read the Full Article

Open access via University of California’s eScholarship repository:

The Random‑Access Image: Memory and the History of the Computer Screen – PDF

It's my 14 year anniversary on Tumblr 🥳

#14 year tumblrversary #tumblr milestone

Gott gebe Glück mit Frieden

#house of glücksburg

Listen to True Companion by Marc Cohn on Apple Music. 1991. Duration: 4:09

https://vm.tiktok.com/ZMJTnEafv/

#davidbowie #bowie #90s #1990s #2000s #00s #music #interview #paxman #internet #prediction #technology #prophecy #social media

Trending Blogs

Recently Viewed Blogs

Robert Boris